ALARM CORRELATION FOR FAULT MANAGEMENT IN TELECOMMUNICATION NETWORKS

Abstract

In communication networks, a large number of alarms are generated to signal any abnormal behavior in the network. A single fault in the network typically results in a large number of alarms. Correlating these alarms to identify faults is a major task in fault management. Alarms that have not been correlated may lead to increased number of alarms (containing both cause and effect alarms) with incomplete and/or redundant information to the user and hence may result in improper diagnosis of faults and increased network outages. Alarm correlation is one of the most important techniques used in telecommunication networks for • real time diagnosis of faults in the network • increasing the quality of information presented in the alarm • reducing the number of insignificant alarms presented to the user In this thesis, the Motorola digital cellular network is studied to understand the problems associated with fault management aspect of network management. The alarm streams generated in the Motorola cellular network, due to the failure of a few important "field replaceable components", are analyzed. The fault management relationship among different devices in the Motorola network is studied and a rule-based alarm correlation engine is implemented to correlate the input alarms to detect faults. The Telecommunication Management Network (TMN) framework is used to report correlated alarms and control the functioning of the correlator.